Pocket-Sized Multimodal AI for content understanding and generation across multilingual texts, images, and ๐ video, up to 5x faster than OpenAI CLIP and LLaVA ๐ผ๏ธ & ๐๏ธ
